Again, I'm not advocating any position. Even against computationalism, the only position I admit to holding, I've not offered an argument. On epiphenomenalism, I've not giving an opinion, just briefly noted some problems with it and alluded to others.
I'm just trying to point out that the problem is incredibly difficult and there are no simple or easy answers. It's silly to make any sort of claim as to the nature of consciousness, let alone what you need to artificially create it! Absurd things like "it must be" type answers are not just uninteresting (they're implicit in the metaphysics, after all, and thus offer us nothing new), they're ultimately useless (they can't get us past "that" to "how").
Again, the problem is extraordinarily complicated. It's absurd to make any claims like "it must be that" or "requires only that".
Right. And since you haven't offered an argument against computationalism, your bold claim that "one thing we can be reasonably certain about, however, is that we can't create anything resembling consciousness by purely algorithmic means" was justifiably attacked. You have some background in philosophy, it seems, but not nearly enough basis to justify your generally smug attitude.
Just for fun:It turns out that this isn't true. See Bell.
Anything that's typically treated as non-deterministic can just as easily be expressed as something with hidden variables;
No. This is an extremely common misconception, but it's still just that. From Wikipedia on Bell's theorem: "one is forced to reject either locality or realism (or both)." This is very different from saying "the universe is provably non-deterministic." As I said, the conclusion that the universe is not goverened by hidden variables rests on assumptions that the hidden variables have locality. It says nothing about hidden variables in a general sense; the universe can still be modeled with hidden variables as long as they don't manifest themselves in certain physical ways.
In fact, this is fairly easy to see: based on empirical data points alone, any model that proposes nondeterminism can be replaced with one that requires no such thing and yields the same data points. All you have to do is replace the source of nondeterminism with invisible-god and his invisible-book-of-random-numbers. As long as invisible-god and his book are not accessible to those collecting the data points, the two are indistinguishable based on those data points.